Download Homeotic genes and the regulation of myoblast migration, fusion

3333
Development 124, 3333-3341 (1997)
Printed in Great Britain © The Company of Biologists Limited 1997
DEV7582
Homeotic genes and the regulation of myoblast migration, fusion, and fibrespecific gene expression during adult myogenesis in Drosophila
Sudipto Roy and K. VijayRaghavan*
National Centre for Biological Sciences, Tata Institute of Fundamental Research, P.O. Box 1234, IISc Campus, Bangalore 560012,
India
*Author for correspondence
SUMMARY
We have investigated the roles of homeotic selector genes in
the migration and fusion of myoblasts, and in the differentiation of adult muscle fibres of Drosophila. Altering
intrinsic homeotic identities of myoblasts does not affect
their segment-specific migration patterns. By transplanting
meso – and metathoracic myoblasts into the abdomen, we
demonstrate that the fusion abilities of myoblasts are independent of their segmental identities. However, transplanted thoracic myoblast nuclei are ‘entrained’ by those
of the host abdominal muscles to which they fuse and are
unable to ‘switch on’ a thoracic muscle-specific reporter
gene. This process is likely to be mediated by homeotic
repression because mis-expression of an abdominal musclespecific homeotic gene, Ultrabithorax, in the thoracic
muscles results in the repression of the thoracic musclespecific reporter gene. Finally, we show that removal of
Ultrabithorax function specifically from muscle cells of the
first abdominal segment, results in the expression of
thoracic muscle properties. Many of these functions of
homeotic genes in muscle patterning in Drosophila could be
conserved during myogenesis in other organisms.
INTRODUCTION
change their properties upon transplantation. However, it is not
possible in this situation to distinguish between autonomous
and non-autonomous roles of Hox genes in specifying their
identity (Olson and Rosenthal, 1994).
Chick/quail chimeras (Ordahl and LeDouarin, 1992) and
grafting experiments with cervical somites (Lance-Jones,
1988) suggest that the environment plays a important role in
muscle patterning. Other experiments, however, suggest
strongly that myoblasts can have heritable, cell-autonomous
information about their positional identity (Donoghue et al.,
1992; Grieshammer et al., 1992; DiMario et al., 1993;
Donoghue and Sanes, 1994).
The molecular mechanisms that operate to specify segmental
identity of muscle fibres and the positional identity of each fibre
within a segment are beginning to be understood in the fruit fly
Drosophila melanogaster (Bate, 1993; Abmayr et al., 1995).
Studies on the fruit fly have demonstrated two important
features: first, the high level of conservation of molecular mechanisms of specification of organ identity and second, the
tremendous manipulative advantage that this organism confers
in addressing questions of general value. While many aspects
of muscle pattern in Drosophila appear to be specified by properties intrinsic to the mesoderm, there is now abundant evidence
that proper muscle formation also requires that the epidermis
and the nervous system send correct inductive signals to pattern
many aspects of muscle development (Hooper, 1986; Lawrence
and Johnston, 1986; Greig and Akam, 1993; Fernandes et al.,
1994; Michelson, 1994; Roy et al., 1997).
Several recent studies have provided clues to the molecular mechanisms that operate during myogenesis (Molkentin and Olson,
1996). The development of muscles to give rise to the precise
pattern observed in the mature animal involves the function of
myogenic regulatory genes common to many muscle types, and
other regulatory elements that act to give each fibre its unique
positional property (Donoghue and Sanes, 1994). In vertebrates,
four basic-helix-loop-helix (bHLH) protein coding genes, myoD,
myogenin, myf5 and mrf4, act during myogenesis to define the
properties of skeletal muscle (Weintraub et al., 1991; Olson and
Klein, 1994). The products of these genes regulate myogenesis
by interacting with members of the myocyte enhancer factor-2
(MEF2) transcription factor family, and activate the expression
of other genes that encode specific structural components of differentiated muscle fibres (Molkentin et al., 1995). In addition, the
bHLH myogenic genes have been shown to regulate the process
of exit from the cell cycle that is required for myogenesis to
proceed (Guo et al., 1995; Halevy et al., 1995). The bHLH genes
are themselves regulated through inductive influences from other
tissues, by mechanisms that appear to be evolutionarily conserved
(Cossu et al., 1996; Molkentin and Olson, 1996).
While these experiments have illuminated the mechanisms
of muscle differentiation, far less is known about how specific
muscle identities are specified. Transplantation experiments
with vertebral precursors have shown that structures that arise
from these precursors are intrinsically determined and do not
Key words: selector gene, myoblasts, migration, fusion,
differentiation, founder cell, Drosophila, Ultrabithorax
3334 S. Roy and K. VijayRaghavan
We have investigated the functions of homeotic genes during
the development of the thoracic and abdominal muscles of the
adult fly (Fig. 1). We have previously shown that during pupal
development, wing disc-associated myoblasts in the second
thoracic segment (T2) migrate out over the pupal epithelium to
reach specific muscle formation sites in the thorax (Fernandes
and VijayRaghavan, 1993; Fernandes et al., 1994). Is this
segment-specific migration process to seek out sites of muscle
formation guided by autonomous properties of myoblasts, or
are they provided with cues from the epidermis? We show that
expression of an abdominal muscle-specific homeotic gene,
Ultrabithorax (Ubx), in the thoracic myoblasts does not affect
the migration pattern of these myoblasts.
In a pioneering experiment, Lawrence and Brower (1982)
showed that myoblasts from the wing imaginal disc, which
normally give rise to the dorsal muscles in T2, can contribute
upon transplantation to diverse muscle types in the adult. This
would suggest that myoblasts can fuse indiscriminately, independent of their segmental identity, and that many properties
of a muscle fibre must be specified by cues external to them.
We have recently shown that myoblasts on the wing disc in
particular, and in T2 in general, do not express any homeotic
selector gene of the Antennapedia (ANTP-C) or bithorax
complex (BX-C) (Roy et al., 1997; Fig. 1). Therefore, one
possible explanation for the results of Lawrence and Brower is
that indiscriminate fusion of wing disc-derived myoblasts was
observed because the donor myoblasts did not have any
intrinsic homeotic identity, since they were of T2 origin. We
therefore tested whether myoblasts expressing a homeotic
selector gene, the Antennapedia (Antp) gene of the ANTP-C
complex, can fuse with abdominal myoblasts that express other
selector genes such as Ubx, abdominal-A (abd-A) or
Abdominal-B (Abd-B). Our results demonstrate that fusion is
independent of myoblast segmental identity.
Although fusion between myoblasts of different segmental
origins can take place, however, the pattern of gene expression
by the donor nuclei in the host syncitium is affected. We show
that transplanted wing disc myoblasts that fuse with abdominal
muscles do not express a thoracic muscle-specific actin
reporter gene, but do so when they migrate back into the thorax
and fuse with developing thoracic muscles. In another experiment, where the abdominal mesodermal selector gene, Ubx, is
ectopically expressed in the developing mesothoracic muscles,
we demonstrate the repression of the thoracic muscle-specific
actin reporter gene. Finally, we examine a situation where the
mesodermal identity of the first abdominal segment (A1) has
been transformed to a thoracic label by removal of Ubx
function specifically in the A1 mesodermal cells. Here we
show that the A1 muscles now express the thoracic musclespecific actin reporter, as a consequence of the removal of Ubx.
We interpret these results with respect to the requirements
for the formation of a mature muscle fibre and discuss possible
mechanisms within the broader question of how muscle development takes place in vertebrates.
MATERIALS AND METHODS
Myoblast transplantation
Wing and haltere discs were isolated from wandering third instar
larvae and the notal region harbouring the imaginal myoblasts
dissected as described in Lawrence and Brower (1982). These disc
fragments were then transplanted into the abdomen of prepupal hosts
as described before (VijayRaghavan et al., 1996). Animals that
survived the transplantation process and eclosed were dissected in
Ringer’s solution and analysed for β-galactosidase activity or GFP
expression, as described below.
Fly strains
The Canton-S strain was used as the wild-type control in all experiments
unless mentioned, and also as a host strain in transplantation experiments. The Myosin Heavy Chain-lacZ (Hess et al., 1989) and the Actin
(88F)-lacZ/Actin (88F)-GFP strains (Hiromi et al., 1986; Barthmaier
and Fyrberg, 1995) were used as donors in transplantation experiments.
Although the Act(88F) gene was initially thought to be indirect-flightmuscle specific, we find that at least the activity of the lacZ reporter gene
Adult
T3 A1
T1 T2
Larva
Expression pattern of homeotic genes in
the somatic mesoderm
Fig. 1. Diagram illustrating the distribution of adult second (T2) and
third thoracic (T3) muscle progenitors in the larva and the types of
adult muscles to which they contribute. Myoblasts on the wing disc
in T2 of the larva (shown in blue) contribute to the large indirect
flight muscles (IFMs) of this segment in the adult (blue muscles in
the adult fly), while those on the haltere disc in T3 (shown in
crimson) develop into T3-specific adult muscles (crimson muscles in
the adult fly). We have investigated the roles of homeotic selector
genes in the migration, fusion and differentiation of these myoblasts
during adult myogenesis. The normal expression patterns of
homeotic selector genes in the somatic muscles in different thoracic
and abdominal segments is colour coded and depicted in segmental
register below the larva. Expression patterns of these genes in the
muscles of the head and terminal abdominal segments have not been
determined with clarity, and hence are not shown. Yellow, Sex combs
reduced (T1); blue, No selector gene expression (T2); crimson,
Antennapedia (T3); green, Ultrabithorax (A1-A5) and brown,
abdominal-A (A4-A7). The expression patterns of homeotic selector
genes shown here has been modified from Bate (1993).
Homeotic control of muscle patterning 3335
muscle formation sites following stereotyped routes. During
from the Act(88F) promoter is seen in all thoracic segments; in specific
muscles in T1, T2 and T3 in addition to the indirect flight muscles in T2
pupal development in Drosophila, adult muscle precursors
(VijayRaghavan et al., 1996). However we have not detected Act(88F)associated with imaginal discs and nerves migrate out and
lacZ or Act(88F)-GFP expression in any abdominal muscle. Thus the
reach muscle formation sites on the developing adult
Act(88F) promoter is a reporter for adult thoracic muscles, though not
epidermis, following well-defined segment-specific patterns. In
all adult thoracic muscles express Act(88F) reporter genes. The UASthe mesothorax, these precursors migrate out in a broad stream
Ubx strain was kindly provided by A. M. Michelson (Brigham and
of cells and reach the flight muscle formation sites on the
Women’s Hospital, Boston, USA). The X-chromosome p-GAL4
developing dorsal notum (Fernandes and VijayRaghavan,
enhancer-trap insertion, 1151, was obtained from L. S. Shashidhara
1993; Fig. 2A). We previously showed that in animals that
(Centre for Cellular and Molecular Biology, Hyderabad, India). The
exhibit a transformation of the metathoracic (T3) epidermis to
expression pattern of this strain was documented using a UAS-lacZ
reporter construct (Brand and Perrimon, 1993). The GAL4 driver
the likeness of T2 due to a combination of regulatory mutations
expresses β-galactosidase from the UAS-lacZ reporter gene in the wing
in the Ubx gene, the number and arrangement of T3 muscle
disc-associated myoblasts, in these myoblasts
during pupal development as they spread over the
developing notum, and in mature adult muscles
Larval muscles
(data not shown). The twist-lacZ strain, which
Everting wing
contains the regulatory domains of the twist gene
fused to the bacterial lacZ, was kindly provided
disc
by F. Perrin-Schmitt and B. Thisse (Strasbourg,
France) and was used to follow the migration
pattern of the myoblasts. For this experiment,
1151; twist-lacZ flies were crossed to UAS-Ubx
flies and the pattern of myoblast migration in the
mesothorax on ectopic Ubx expression was
Migrating myoblasts
followed using twist-lacZ expression in the
pupae of the progeny.
Histochemistry and
immunocytochemistry
Histochemical staining for β-galactosidase
activity was done according to standard procedures. For analysis of both β-galactosidase
and GFP expression, muscle fibres were first
briefly fixed in 4% paraformaldehyde in
phosphate buffered saline (PBS), lightly
stained for β-galactosidase expression till the
appearance of blue colour, washed thoroughly
with several changes of PBS and then
analysed for GFP expression using fluorescence optics (see below). The anti-Ubx
monoclonal antibody FP 3.38 was used at a
dilution of 1:50 and was a generous gift of R.
White (Cambridge University, UK). The
antibody reaction was developed using the
Vectastain ABC Elite kit (Vector Labs, USA)
according to the manufacturer’s instructions.
Microscopy
Tissue preparations were either mounted
directly in GelTol (Immunon, USA) or 50%
glycerol, or dehydrated through ethanol grades,
cleared in methyl salicylate and mounted in
Canada Balsam. Preparations were viewed
using polarised light (unstained muscles),
Nomarski optics (stained muscles, imaginal
discs and pupal tissue) or epifluorescence
optics (for GFP expression in muscles) using
340-380 nm exciting light.
RESULTS
Homeotic genes do not specify
migration patterns of myoblasts
During development, myoblasts migrate to
Fig. 2. Mis-expression of Ubx, an abdominal muscle identity homeotic gene, in the
myoblasts of T2 does not alter their migration pattern. (A) Diagram showing the migration
pattern of myoblasts in T2 during early pupal development. The three larval muscles
indicated escape histolysis during pupation, and serve as templates for the development of
one set of IFMs, the dorsal longitudinal muscles (DLMs). (For further details on the wildtype pattern of myoblast migration see Fig. 8 in Fernandes and VijayRaghavan (1993)).
(B) Ubx expression in the myoblasts associated with the wing disc in T2 (arrowheads) from
the UAS-Ubx transgene under the influence of the 1151 GAL4 driver. This pattern of
myoblasts on the wing disc is similar to the wild-type pattern (see Fig. 3 in Fernandes et al.
(1994) for wild-type pattern of myoblasts on the wing disc). (C) Migration pattern of Ubxexpressing (see Materials and methods) myoblasts in T2 revealed by expression of the twistlacZ reporter gene (blue staining), which is expressed in all undifferentiated myoblasts. The
diffuse staining is due to the cytoplasmic localisation of the β-galactosidase product in the
myoblasts. The filled asterisk marks the position where one set of IFMs, the dorsoventral
muscles (DVMs), will develop. The open asterisks mark the positions of the three remnant
larval muscles that serve as templates for the development of the DLMs. The myoblasts are
seen over the larval templates. This pattern is indistinguishable from wild type (compare
with A, and Fig. 8 in Fernandes and VijayRaghavan (1993)).
3336 S. Roy and K. VijayRaghavan
precursors and their migration pattern during pupation is transformed to that of a T2 identity (Fernandes et al., 1994). This
indicated that the process of myoblast migration is very likely
determined by cues provided by the epidermis and is independent of the segmental identity of the migrating myoblasts themselves. In order to further clarify the role of myoblast identity
in their migration process, we mis-expressed an abdominal
muscle-specific selector gene Ubx using a GAL4 line called
1151 and the UAS-Ubx transgene in the wing disc-associated
T2 myoblasts, and determined its effect on the migration of
these cells during pupal development. We analysed the
expression pattern of 1151 by crossing it to the UAS-lacZ
reporter strain (Brand and Perrimon, 1993) and find that it is
expressed in all imaginal disc and nerve-associated adult
muscle precursors, and also during their differentiation and in
almost all mature adult muscles (see Materials and methods).
When Ubx is expressed with this GAL4 driver, we find that T2
myoblast migration pattern is unaffected and that these
myoblasts are able to follow defined routes to reach the sites
of muscle formation (Fig. 2B,C). This observation provides
clear evidence that the intrinsic identity of the myoblasts is
irrelevant in determining their segment-specific migration
patterns, which are very likely determined entirely by the
segmental identity of the epidermis.
Homeotic genes do not specify fusion properties of
myoblasts
We transplanted the presumptive notum region of wing
imaginal discs, harbouring myoblasts that give rise to the
indirect flight muscles (IFMs) of T2, into the abdomen of
prepupal hosts (Fig. 3A). Donor myoblasts were from the
Myosin Heavy Chain-lacZ (MHC-lacZ) transformant strain and
hosts were wild type. We used myoblasts from the MHC-lacZ
strain because this transgene is expressed in all differentiated
muscle fibres in Drosophila and hence would serve as a useful
marker to follow the fates of the transplanted myoblasts (Hess
et al., 1989). We find that myoblasts from the wing discs can
fuse with the developing abdominal muscles of the host (Fig.
3B). The myoblasts on the wing imaginal discs will normally
contribute to the dorsal muscles of the adult mesothorax
(Lawrence, 1982; Fig. 1). The transplantation experiments
show that these myoblasts, when freed of the disc epithelium,
can contribute to diverse muscle types. This result essentially
reconfirmed Lawrence and Brower’s previous observation
using wild-type wing disc-derived donor myoblasts transplanted into hosts that were mutant for a gene that affected the
activity of the enzyme succinate dehydrogenase (Lawrence and
Brower, 1982). However, our recent study on homeotic gene
expression and function in the thoracic mesoderm has revealed
that no known homeotic selector gene of the ANT-C or the BXC is expressed or required for the specification of muscle
pattern in T2 in either the embryo or the adult. Specifically,
myoblasts associated with the wing imaginal disc in this
segment do not express any selector gene (Roy et al., 1997).
The same study also revealed that muscle pattern in T3 is
specified by the activity of the selector gene Antp, and
myoblasts associated with the haltere imaginal discs of this
segment, unlike those on the wing disc in T2, express Antp.
We reasoned that the ability of the wing disc myoblasts to
indiscriminately fuse and contribute to abdominal muscles may
arise from the fact that these cells are not committed by the
activity of any selector gene. We therefore transplanted
portions of haltere discs that harbour the T3-specific myoblasts
from MHC-lacZ donors into the abdomen of wild-type hosts
(Fig. 3A). We predicted that if restrictions to fusion in these
myoblasts are dictated by Antp expression, then unlike the
wing disc-derived myoblasts, these cells will not be able to fuse
with abdominal myoblasts and contribute to abdominal
muscles. However on transplantation, haltere disc myoblasts
behave in a very similar way to wing disc myoblasts and can
fuse with and contribute to host abdominal muscles (Fig. 3C).
These results suggest that imaginal myoblasts are not restricted
by segment-specific fusion properties through the activity of
homeotic genes, but are indeed quite promiscuous in their
ability to participate in diverse myogenic programmes.
Repression of segment-specific gene expression in
thoracic myoblast nuclei that fuse with abdominal
fibres
Though thoracic imaginal myoblasts appear not to exhibit any
restrictions to their fusion capabilities, it is possible that their
nuclei are able to maintain a thoracic identity even on being
part of an abdominal muscle. In vertebrates for instance, there
is evidence that the patterns of gene activity in the nuclei of a
muscle fibre can indeed be very diverse (Sanes et al., 1991).
Alternatively, the transcriptional status of the donor thoracic
nuclei could be controlled by the activity of selector genes
expressed in nuclei of the abdominal myoblasts and be
‘entrained’ by them. We next examined whether thoracic
muscle nuclei are able to express thorax-specific differentiation
genes even while coexisting with abdominal muscle nuclei or
whether they get ‘entrained’ to execute an abdominal musclespecific differentiation programme.
We transplanted wing disc myoblasts doubly marked with
two different transgenes: MHC-lacZ and Actin (88F)-Green
fluorescent protein (Act(88F)-GFP). The Actin (88F) gene is
expressed in thoracic muscles, predominantly in the IFM fibres
(Hiromi et al., 1986; Barthmaier and Fyrberg, 1995; See also
Materials and methods). Fig. 4A shows the expression of GFP
in the IFMs of an Act(88F)-GFP fly. In the transplantation
experiments, MHC-lacZ expression monitored the fusion of
myoblasts and the Act(88F)-GFP expression would test the
ability of these myoblasts to express a thorax-specific gene. We
find that while wing disc myoblasts are able to fuse and contribute to abdominal muscles, as observed by the expression of
the MHC-lacZ transgene, they are unable to express the
Act(88F)-GFP reporter gene, suggesting that on fusion their
nuclei are ‘entrained’ by those of the abdominal muscles to
prevent Act(88F) gene expression (Fig. 4B,C). Since we could
observe GFP fluorescence in flight muscles of Act(88F)-GFP;
MHC-lacZ flies even after fixation and β-galactosidase staining
(see Materials and methods), the absence of GFP fluorescence
in the abdominal muscles in the transplantation experiments is
unlikely to be due to quenching effects from fixation and βgalactosidase staining.
We have done similar transplantation experiments with wing
disc-derived myoblasts marked with the Act(88F)-lacZ
transgene and, in every case examined, we were unable to
detect any β-galactosidase expression from the Act(88F)
reporter in the abdominal muscles, though in these experiments, unlike the one above, for want of a suitable marker, we
were unable to monitor fusion events of the thoracic myoblasts
Homeotic control of muscle patterning 3337
with the abdominal muscles. However, the sensitiveness of this
latter set of experiments is borne out by the fact that in cases
where transplantations with Act(88F)-lacZ-marked wing disc
myoblasts occurred close to T2, i.e. on the first or second
abdominal segments (A1 and A2), some of the donor
myoblasts could migrate back into T2 and fuse with the developing IFMs in that segment. In such cases, we were able to
observe Act(88F)-lacZ expression even from single transplanted myoblast nuclei that happened to fuse with the developing IFMs. One such example is shown in Fig. 4D.
The inability of thoracic myoblast nuclei to express the
Act(88F) reporter genes in the abdominal muscles indicates
that the expression of this gene is a segmental property of the
thoracic muscles, and that the thoracic myoblasts on fusion
with abdominal muscles are ‘entrained’ into an abdominal
identity, most likely by the activity of abdominal musclespecific selector genes such as Ubx, abd-A and Abd-B (Bate,
1993; also see below).
Ubx mis-expression in the developing flight muscles
represses the expression of the thoracic musclespecific reporter gene, Act(88F)-lacZ
To test if any of the homeotic genes that are normally expressed
in the abdominal muscles are negative regulators of the
Act(88F) gene, Ubx was expressed in the developing IFMs
using the GAL4/UAS system (Brand and Perrimon, 1993). For
this mis-expression study, the GAL4 driver 1151, which was
previously used to express Ubx in the T2 myoblasts to
determine the effect of this mis-expression on myoblast
migration, was used. As mentioned earlier, this GAL4 driver
was found to be expressed in the differentiating myoblasts as
well as in most mature adult muscles. During normal development, expression of the Act(88F)-lacZ transgene is first seen at
about 14-16 hours after puparium formation (APF) in the differentiating IFM fibres (Fig. 5A). Expression of Ubx in these
developing muscles resulted in the repression of Act(88F)-lacZ
expression (Fig. 5B). Almost no expression of this reporter
gene was seen, and the indirect flight muscles failed to
complete development and eventually degenerated. The
absence of Act(88F)-lacZ expression cannot be attributed to the
fact that Ubx expression in T2 myoblasts resulted in inappropriate cell death or inhibited myoblast differentiation, because
the myoblasts were able to migrate to IFM formation sites and
fuse to form developing myofibres, and at a stage in development when these myofibres should normally be expressing the
Act(88F) reporter, the expression of this transgene was not
observed in these muscle fibres (Fig. 5B). Thus, Ubx misexpression in T2 myoblasts resulted in the alteration of their
identity into that of an abdominal one and these myoblasts
therefore were unable to execute a differentiation programme
that is appropriate to that of the thoracic segment.
Removal of Ubx expression from abdominal
muscles allows the expression of Act(88F)-lacZ
In animals homozygous for anterobithorax (abx), bithorax3
(bx3), postbithorax (pbx), three mutant alleles of the Ubx gene
of the BX-C, a transformation of the epidermis of T3 towards
that of T2 is observed. Such adults have two pairs of wings. In
our analysis of flight muscle development in animals carrying
the above ‘triple mutation’ combination in Ubx, we also
demonstrated that the mesoderm of the first abdominal segment
took on properties of the third thoracic segment (Fernandes et
al., 1994). Thus, while the ectoderm was transformed from T3
towards T2, the mesoderm was transformed from A1 towards
T3. We also showed that in the wild-type animals, Ubx was
expressed at its most anterior mesodermal domain in A1. In
‘triple mutant’ animals, Ubx expression in A1 mesoderm is
absent (Fernandes et al., 1994). High levels of Antp expression
in the mesoderm are seen, at its most anterior domain in T3,
and the level is significantly lower in A1 mesoderm of the wildtype. In the ‘triple mutant’, high levels of Antp expression are
also seen in the A1 mesoderm (Fernandes et al., 1994). On the
basis of the transformation of the mesoderm observed in A1 of
the ‘triple mutant’, the absence of transformation in the
ectoderm of this segment, and the pattern of expression of Ubx
and Antp in the mesoderm of wild type and mutant animals,
we concluded that Ubx expression in the mesoderm of A1 is
autonomously required for proper muscle development and
that absence of Ubx expression leads to the transformation
observed. We examined A1 muscles in wild type and ‘triple
mutant’ animals for the expression of the Act(88F)-lacZ
reporter. While expression of this reporter was never observed
in wild-type A1 muscles, it was found that absence of Ubx in
A1 mesoderm of the ‘triple mutant’ leads to defects in muscle
patterning, and allows the expression of Act(88F)-lacZ in A1
muscles (Fig. 5C,D). Thus, while expression of Ubx in T2
mesoderm alters its identity into that of abdominal one and
modifies its differentiation programme such that it is no longer
able to express thorax-specific differentiation genes, removal
of Ubx expression in A1 mesoderm results in the alteration of
the identity of the A1 muscles into that of a thoracic label, and
these muscles now follow a thoracic muscle differentiation
programme and express thoracic muscle-specific genes.
DISCUSSION
We have addressed specific questions regarding the role of
homeotic selector genes during muscle development. In the
epidermis of Drosophila, where their function has been best
investigated, these genes function cell-autonomously to specify
the developmental identity of parasegments (Lawrence and
Morata, 1994). Not only are the selector genes involved in
promoting a unique pattern of differentiation of the cells that
belong to a particular parasegment, but they are also involved
in regulating their mixing properties and in defining cellular
affinities (Lawrence, 1992). While there is evidence that
homeotic genes do function autonomously in muscle cells to
specify pattern, their roles in different aspects of myogenesis
remain obscure. It would be reasonable, for example, to expect
that as in the epidermis, homeotic genes regulate the surface
properties of mesodermal cells and thereby channel them to
migrate along defined paths and to participate in specific fusion
programmes. However, our present study indicates that these
properties of myoblasts are not regulated by the autonomous
function of selector genes in these cells; a situation very
different from that in the epidermis. This difference in the
activity of homeotic genes in the two germ layers may have to
do with the peculiar way by which muscles develop (Lawrence,
1992). It is now a well established fact, at least in the
Drosophila embryo, that each muscle fibre is prefigured by a
single cell called a ‘founder cell’. One extreme view of the role
3338 S. Roy and K. VijayRaghavan
of ‘founder cells’ is that their segmental identity and positional
information are necessary and sufficient to pattern the
formation of a muscle fibre. Other myoblasts that fuse with the
‘founder’ are ‘feeder myoblasts’ of no distinct identity, and
they will be submissive, in their pattern of gene expression, to
Wing disc
Haltere disc
Myoblasts
T2
T1
T3
controls by regulatory proteins made by the ‘founder cell’
nucleus. In this view, myoblasts from different segments will
be able to fuse with each other, but the developing muscle fibre
will take on properties decided by the ‘founder cell’
(Lawrence, 1985). An alternative extreme view could be that
fibre properties are a mix of the properties of its constituent
nuclei, with no nucleus being ‘more equal’. In this scheme, the
property of a fibre will be determined by the identity of the
majority of its nuclei. While studies from vertebrates have
suggested that muscle nuclei have differential abilities to
express synapse-specific genes (Sanes et al., 1991), we do not
know if properties such as the segmental identity of a fibre are
determined by the activity of one or many nuclei. Similarly,
muscle ‘founder cells’, which have only one nucleus, send out
processes that attach to specific epidermal attachment sites
(Rushton et al., 1995). Is the ability to recognise these
epidermal attachment sites a result of cues from the epidermis,
A1
Prepupa
Fig. 3. When transplanted into the abdomen, thoracic myoblasts,
irrespective of their homeotic identities, are able to fuse with
abdominal myoblasts and form abdominal muscles. (A) Diagram
illustrating the transplantation technique that was used to determine
the fusion abilities of myoblasts. The notal regions of wing and
haltere discs harbouring the myoblasts were dissected (the straight
line across the discs represents the line of dissection) and
transplanted into the abdominal regions of prepupal hosts. The
myoblasts on the wing disc do not express any homeotic selector
gene (represented in blue), while those associated with the haltere
disc express Antp (represented in crimson). (B) Dorsal muscles in the
fifth abdominal segment of a host fly that arose from the fusion of
wing disc-derived T2-specific myoblasts (marked with the MHClacZ transgene) with abdominal myoblasts (arrowheads). Muscles
which developed without fusion of MHC-lacZ-marked T2 myoblasts
are unstained for β-galactosidase expression and are white.
(C) Pleural muscles in the fifth abdominal segment of a host fly that
arose from the fusion of haltere disc derived T3-specific myoblasts
(marked as in above with the MHC-lacZ transgene) with abdominal
myoblasts (arrowhead). The neighbouring unstained muscles have
arisen only from the abdominal myoblasts and hence do not show βgalactosidase expression (asterisks). For pattern of adult abdominal
muscles see Bate (1993).
Fig. 4. On fusion, thoracic myoblast nuclei lose their segmentspecific identities and are entrained by those of the host muscle.
(A) Act(88F)-GFP expression in the IFMs of an adult fly on
excitation with 340-380 nm light. The asterisks mark the six dorsal
longitudinal muscle fibres (DLMs) and the arrowheads indicate two
groups of dorsoventral muscles (DVMs). In this figure, anterior is to
the right. (B) Muscles in the fourth abdominal segment that have
developed by the fusion of T2 myoblasts from the wing disc (marked
with MHC-lacZ and Act(88F)-GFP) with abdominal myoblasts (blue
muscles marked with arrowheads). (C) The same muscles shown in
B on excitation with 340-380 nm light do not show GFP fluorescence
(arrowhead; compare with GFP expression in A). (D) β-galactosidase
expression in a donor T2 myoblast nucleus (arrow) from the
Act(88F)-lacZ strain that had migrated into the thorax from the
abdomen (the site of transplantation) and had fused with the
developing IFMs in T2.
Homeotic control of muscle patterning 3339
Myoblasts
Mature muscle
Developing fibre
A
Fig. 5. Mis-expression of Ubx in the developing thoracic
muscles results in the repression of thoracic musclespecific Act(88F)-lacZ reporter gene, while loss of Ubx
expression from abdominal muscles results in the
activation of this reporter in these muscles. (A) A 24hour APF wild-type pupal preparation, showing the
expression pattern of Act(88F)-lacZ in the developing
IFMs. The arrowheads indicate two developing DLM
fibres while the asterisks mark two groups of developing
DVMs. (B) Act(88F)-lacZ expression is not observed on
mis-expression of Ubx in the developing IFMs at 24
hours APF. Note the extremely low level of βgalactosidase staining in one developing DLM fibre
(arrowhead). The asterisk marks the position of a
developing DVM. Note the complete absence of βgalactosidase expression in this muscle. The developing
IFMs in this preparation are shown at a higher
magnification than in A. (C) Polarised light picture of
muscle fibres (arrowheads) in the A1 segment of an
Act(88F)-lacZ transformant fly. Expression of the
Act(88F)-lacZ transgene is normally not observed in
these muscles. (D) Polarised light picture of A1 muscle
fibres in a ‘triple mutant’ animal. Absence of Ubx
function in A1 mesodermal cells of these mutants
results the transformation of their identity to that of T3
mesoderm. This results in defects in muscle patterning,
and allows the expression of the thoracic musclespecific Act(88F)-lacZ transgene (arrowheads). Not all
T3 muscles normally express the Act(88F) reporter,
therefore lacZ expression is not observed in all
homeotically transformed A1 muscles in the ‘triple
mutant’ (See Materials and methods).
neuron
synapse
epidermis
B
C
Fig. 6. Diagram illustrating the relative importance of the identities
of the epidermis, the motor neuron, the ‘founder cell’ and the
contributing myoblasts for the elaboration of segment-specific
muscle patterns. (A) The development of a mature muscle fibre with
proper attachment and innervation requires that the identities of the
‘founder cell’, the epidermis that it attaches to and the motor neuron
that innervates it, are compatible. The ‘founder’ is represented by the
oblong cell while the other myoblasts are depicted as circles.
(B) When donor myoblasts from one segment (shown in red) are
transplanted to the sites of muscle formation in another segment,
these myoblasts are able to fuse with the resident host myoblasts of
that segment (shown in blue) and form a mature fibre, as long as the
‘founder’ is of the right segmental identity (represented by the blue
oblong cell). Even in the presence of a vast majority of donor
myoblasts, the ‘founder’ will be able to ‘entrain’ them to its pattern
of gene expression and make a wild-type muscle fibre. (C) However,
if a ‘founder’ from the donor myoblast population, and hence of a
wrong segmental identity (represented by the red oblong cell),
nucleates the process of muscle development, it will not be able to
recognise the proper attachment sites on the mismatched epidermis,
nor will it be able to direct the formation of a proper pattern of
innervation. This will result in derangement or degeneration of the
muscle fibre.
3340 S. Roy and K. VijayRaghavan
or are factors intrinsic to each muscle also involved? If the
latter, are the muscle-derived components specified by the
‘founder’ nucleus, with the later arriving myoblast nuclei
having no effect on attachment? Analysis of mutant embryos
that are specifically affected in myoblast fusion has revealed
that in the absence of ‘feeder myoblast’ fusion, the ‘founder
cell’ is nevertheless able to recognise its correct attachment
sites, synthesise contractile proteins and get appropriately
innervated, suggesting that these cells do indeed have all the
information required to generate the final muscle pattern
(Rushton et al., 1995). However, it remains to be seen whether
during normal development such a mechanism is operative, or
whether other contributing myoblasts also exert an influence
on the development and identity of the mature muscle fibre.
While it is not known whether the ‘founder cell’ mechanism
operates during adult muscle development in Drosophila,
many features of myogenesis suggest that differences between
myoblasts exist and are required for normal muscle development. It is possible that imaginal disc-associated myoblasts that
we and others have used in transplantation assays represent the
uncommitted ‘feeder myoblasts’ we have alluded to above. As
in the embryo, ‘founder cells’ could be specified at a certain
developmental stage during adult myogenesis, from among
these cells, by mechanisms of cell-cell interactions together
with extrinsic cues provided by interacting tissues (Bate et al.,
1993; Baylies et al., 1995). It is also possible that ‘founders’
may be present among other groups of myoblasts such as those
associated with the larval nerves, and they may also take the
form of structures like persistent larval templates, as in the case
of the developing dorsal longitudinal muscles in T2 (Fernandes
et al., 1991) Our recent studies on the expression patterns of
an even skipped-lacZ (eve-lacZ) transgene has revealed that it
is expressed in subsets of nerve-associated adult myoblasts in
the larva, but not in wing disc-associated myoblasts, making
eve a possible candidate for a ‘founder cell’-specific marker
gene for adult muscles (S.R., unpublished). In keeping with
this observation is the fact that eve is also expressed in the
‘founders’ of subsets of larval muscles during embryonic
development (Bate, 1993). The progenitor of the ventral adult
muscles in each abdominal segment has been identified, and it
arises from a single ‘founder-like’ cell in the embryo (Carmena
et al., 1995). The formation of adult fibres by descendants of
this cell raises several interesting questions about the relative
potential of sibling myoblasts to configure muscle pattern, and
about how and when differences, if any, amongst these
myoblasts could arise.
The essence of the above argument on the roles of ‘founders’
in muscle pattern specification is schematically summarised in
Fig. 6. A ‘founder’ myoblast is selected from a pool of
myoblasts. This ‘founder’ cell sends out filopodia to epidermal
attachment sites, which will later develop to form the
apodemes. Other myoblasts fuse to the ‘founder cell’ and contribute to the growth of a mature muscle fibre. Developing
neurons innervate the fibre and form a defined pattern of
synaptic boutons. Thus, for a mature muscle to be observed in
the adult, the ‘founder cell’ must recognise correct attachment
sites on the epidermis, i.e. muscle and epidermis must have the
same ‘segmental address’ and the developing innervation must
recognise the muscle fibre as being of the same ‘address’ (Fig.
6A). When myoblasts from another segment are ectopically
transplanted to the site of a developing fibre, they fuse with the
fibre and can contribute to the muscle that is formed. A mature
fibre with a heterogeneous population of donor and host nuclei
is observed only when the ‘founder cell’ belongs to the host.
Only then will the appropriate epidermal and neuronal
‘addresses’ be read. A mature fibre can develop even when the
majority of the myoblasts are from another segment
(Lawrence, 1985). As long as the ‘founder cell’ is of the right
segmental identity, it will be able to ‘entrain’ the nuclei of other
myoblasts, resulting in the development of a mature fibre (Fig.
6B). However, if the ‘founder cell’ is formed by the contribution from the ‘wrong’ segment, as in Fig. 6C, we suggest that
a mature muscle is unlikely to form even if a substantial
number of the myoblasts are of the correct segmental identity.
In surface transplant experiments of whole thoracic discs,
VijayRaghavan et al. (1996) found rare situations where
clumps of transplanted muscle were associated with the
abdominal epidermis but expressed thoracic markers. These
muscles were clearly not part of the normal pattern of
abdominal musculature. While it is not possible to say how
many, if any, abdominal myoblasts contributed to these fibres,
such clumps of fibres could represent aborted muscles with a
‘wrong’ segmental identity.
Finally, what do these lessons learnt from myogenesis in
flies tell us about muscle patterning in other organisms,
specially vertebrates? As mentioned earlier, large gaps remain
in our understanding of skeletal muscle patterning during vertebrate embryogenesis. While recent work has indicated that
myoblasts in vertebrates are very diverse, much remains to be
understood regarding the mechanisms of how diversity is
generated, and how muscle patterns arise from the interplay
between intrinsic properties of myoblasts and influences
extrinsic to them (Miller, 1992). Several interesting similarities
are readily apparent when our results are compared with some
elegant studies of a similar nature that have been done in vertebrates. For instance, transplantation of rodent satellite
myoblasts during postnatal development have shown that they
can indiscriminately fuse to many different fibre types and take
on the myosin heavy chain isoform expression patterns of the
fibre to which they have fused (Hughes and Blau, 1992). This
would suggest that satellite cells are similar to Drosophila
‘feeder myoblasts’ in that they are not committed to participate
in any specific myogenic program, and that their gene
expression patterns can be ‘entrained’ by the nuclei of the fibre
to which they happen to fuse. It is possible that mammalian
and avian primary myoblasts and pioneer myoblasts in teleosts
are similar to ‘founder myoblasts’ in Drosophila and other
insects, and it will be interesting to investigate the flexibility
of their developmental potentials using similar experimental
paradigms. In fact, transplantation experiments with clones of
primary myoblasts and satellite cells in birds have demonstrated that fates of muscle fibres are stringently controlled by
myogenic programs intrinsic to these myoblasts during early
development (DiMario et al., 1993). This would suggest that
cells that are involved in the earliest stages of muscle pattern
formation are indeed much less permissive as far as their developmental potentials are concerned, and once the pattern has
been seeded by these cells, other myoblasts which fuse with
them, do exhibit a considerable degree of developmental flexibility. Like many other molecular functions that have been
shown to be conserved between vertebrates and lower animals,
we would like to think that much of what we have gleaned from
Homeotic control of muscle patterning 3341
our experiments on the functions of homeotic genes in diverse
aspects of myogenesis in Drosophila will also reflect the
manner in which their homologs, the Hox genes, would
function during pattern formation in the somatic mesoderm of
vertebrates.
We thank E. Fyrberg, A. M. Michelson and L. S. Shashidhara for
fly strains, R. White for antibodies and J. Sanes, P. Lawrence, M. Bate
and V. Rodrigues for insightful comments on the manuscript. This
work was supported by grants from the Human Frontier Science
Program and from the National Centre for Biological Sciences to K.
V. R.
REFERENCES
Abmayr, S. M., Erickson, M. S. and Bour, B. A. (1995). Embryonic
development of larval body wall musculature of Drosophila melanogaster.
Trends Genet. 11, 153-159.
Barthmaier, P. and Fyrberg, E. (1995). Monitoring development and
pathology of Drosophila indirect flight muscles using green fluorescent
protein. Dev. Biol. 169, 770-774.
Bate, M. (1993). The mesoderm and its derivatives. In The Development of
Drosophila melanogaster, Vol. 2 (ed. M. Bate and A. Martinez Arias), pp.
1013-1090. Cold Spring Harbor Laboratory Press, New York.
Bate, M., Rushton, E. and Frasch, M. (1993). A dual requirement for
neurogenic genes in Drosophila myogenesis. Development Supplement
149-161.
Baylies, M. K., Martinez Arias, A. and Bate, M. (1995). wingless is required
for the formation of a subset of muscle founder cells during Drosophila
embryogenesis. Development 121, 3829-3837.
Brand, A. and Perrimon, N. (1993). Targeted gene expression as a means of
altering cell fates and generating dominant phenotypes. Development 118,
401-415.
Carmena, A., Bate, M. and Jimenez, F. (1995). lethal of scute, a proneural
gene, participates in the specification of muscle progenitors during
Drosophila embryogenesis. Genes Dev. 9, 2373-2383.
Cossu, G., Tajbakhsh, S. and Buckingham, M. (1996). How is myogenesis
initiated in the embryo? Trends Genet. 12, 218-223.
DiMario, J. X., Fernyak, S. E. and Stockdale, F. E. (1993). Myoblasts
transferred to the limbs of embryos are committed to specific fates. Nature
362, 165-167.
Donoghue, M. J., Morris-Valero, R., Johnson, Y. R., Merlie, J. P. and Sanes,
J. R. (1992). Mammalian muscle cells bear a cell-autonomous, heritable
memory of their rostrocaudal position. Cell 69, 67-77.
Donoghue, M. J. and Sanes, J. R. (1994). All muscles are not created equal.
Trends Genet. 10, 396-401.
Fernandes, J., Bate, M. and VijayRaghavan, K. (1991). The development of
the indirect flight muscles of Drosophila. Development 113, 67-77.
Fernandes, J. and VijayRaghavan, K. (1993). The development of indirect
flight muscle innervation in Drosophila melanogaster. Development 118,
215-227.
Fernandes, J., Celniker, S. E., Lewis, E. B. and VijayRaghavan, K. (1994).
Muscle development in the four-winged Drosophila and the role of the
Ultrabithorax gene. Curr. Biol. 4, 957-964.
Greig, S. and Akam, M. (1993). Homeotic genes autonomously specify one
aspect of pattern in the Drosophila mesoderm. Nature 362, 630-632.
Grieshammer, U., Sassoon, D. and Rosenthal, N. (1992). A transgene target
for positional regulators marks early rostrocaudal specification of myogenic
lineages. Cell 69, 79-93.
Guo, K., Wang, J., Andres, V., Smith, R. C. and K. Walsh. (1995). MyoDinduced expression of p21 inhibits cyclin-dependent kinase activity upon
myocyte terminal differentiation. Mol. Cell. Biol. 15, 3823-3829.
Halevy, O., Novitch, B. G., Spicer, D. B., Skapek, S. X., Rhee, J., Hannon,
G., Beach, D. and Lassar, A. B. (1995). Correlation of terminal cell cycle
arrest of skeletal muscle with induction of p21 by MyoD. Science 267, 10181021.
Hess, N., Kronert, W. A. and Bernstein, S. I. (1989). Transcriptional and posttranscriptional regulation of Drosophila myosin heavy chain expression. In
Cellular and Molecular Biology of Muscle Development (ed. L. H. Kedes and
F. E. Stockdale), pp. 621-631. A. R. Liss Inc. New York.
Hiromi, Y., Okamoto, H., Gehring, W. J. and Hotta, Y. (1986). Germ line
transformation with Drosophila mutant actin genes induces constitutive
expression of heat shock genes. Cell 44, 293-301.
Hooper, J. (1986). Homeotic gene expression in muscles of Drosophila larvae.
EMBO J. 5, 2321-2329.
Hughes, S. and Blau, H. M. (1992). Muscle fibre pattern is independent of
lineage in postnatal rodent development. Cell 68, 659-671.
Lance-Jones, C. (1988). The effect of somite manipulation on the development
of motoneuron projection patterns in the embryonic chick hind limbs. Dev.
Biol 126, 408-419.
Lawrence, P. A. (1982). Cell lineage of the thoracic muscles of Drosophila.
Cell 29, 493-503.
Lawrence, P. A. and Brower, D. L. (1982). Myoblasts from Drosophila wing
discs can contribute to developing muscles throughout the fly. Nature 295,
55-57.
Lawrence, P. A. (1985). Pattern formation in the muscles of Drosophila. Cold
Spring Harbor Symp. Quant. Biol. 1, 113-117.
Lawrence, P. A. and Johnston, P. (1986). The muscle pattern of a segment of
Drosophila may be determined by neurons and not by contributing
myoblasts. Cell 45, 505-513.
Lawrence, P. A. (1992). The Making of a Fly: The Genetics of Animal Design.
Blackwell Scientific Publications, Oxford.
Lawrence, P. A. and Morata, G. (1994). Homeobox genes: Their function in
Drosophila segmentation and pattern formation. Cell 78, 181-189.
Miller, J. B. (1992). Myoblast diversity in skeletal myogenesis: How much and
to what end? Cell 69, 1-3.
Michelson, A. M. (1994). Muscle pattern diversification in Drosophila is
determined by the autonomous function of homeotic genes in the embryonic
mesoderm. Development 120, 755-768.
Molkentin, J. D., Black, B. L., Martin, J. F. and Olson, E. N. (1995). Cooperative activation of muscle gene expression by MEF2 and myogenic
bHLH proteins. Cell 172, 19-32.
Molkentin, J. D. and Olson, E. N. (1996). Defining the regulatory networks for
muscle development. Curr. Opin. Genet. Dev. 6, 445-453.
Olson, E. N. and Rosenthal, N. (1994). Homeobox genes and muscle
patterning. Cell 79, 9-12.
Olson, E. N. and Klein, W. H. (1994). bHLH factors in muscle development:
deadlines and commitments, what to leave in and what to leave out. Genes
Dev. 8, 1-8.
Ordahl, C. P. and LeDouarin, N. M. (1992). Two myogenic lineages within
the developing somite. Development 114, 339-353.
Roy, S., Shashidhara, L. S. and VijayRaghavan, K. (1997). Muscles in the
Drosophila second thoracic segment are patterned independent of
autonomous homeotic gene function. Curr. Biol. 7, 222-227.
Rushton, E., Drysdale, R., Abmayr, S. M., Michelson, A. M. and Bate, M.
(1995). Mutations in a novel gene, myoblast city, provide evidence in support
of the founder cell hypothesis for Drosophila muscle development.
Development 121, 1979-1988.
Sanes, J. R., Johnson, Y. R., Kotzbauer, P. T., Mudd, J., Hanley, T.,
Martinou, J-C. and Merlie, J. P. (1991). Selective expression of an
acetylcholine receptor-lacZ transgene in synaptic nuclei of adult muscle
fibres. Development 113, 1181-1191.
VijayRaghavan, K., Gendre, N. and Stocker, R. (1996). Transplanted wing
and leg imaginal discs in Drosophila melanogaster demonstrate interactions
between epidermis and myoblasts in muscle formation. Dev. Genes Evol.
206, 46-53.
Weintraub, H., Davis, R., Tapscott, S., Thayer, M., Krause, M., Benezra,
R., Blackwell, T. K., Turner, D., Rupp, R., Hollenberg, S., Zhuang, Y. and
Lassar, A. (1991). The myoD gene family: nodal point during specification
of muscle cell lineage. Science 251, 761-766.
(Accepted 2 July 1997)